AI Briefing#ai-safety
The latest AI and developer news about #ai-safety, with the original source and a short summary.
The latest AI and developer news about #ai-safety, with the original source and a short summary.
The Anthropic Mythos preview was accessed without authorization in a third-party vendor environment, fueling controversy over exaggerated claims.
It summarizes GPT-5.5's scope of use, safety evaluations, and new guardrails.
OpenAI has released an open-weight model that erases PII such as names and account numbers.
Anthropic has recruited the former Novartis CEO, moving to secure expertise in AI regulation and biotech.
Sam Altman criticized Anthropic's Mythos as fear-based marketing.
There was unauthorized access to Anthropic's Mythos, and some users used it even before the official testing began.
A Gallup survey found that low-income Americans stand out for relying on AI for health information and for their distrust of the medical system.
NSA is using Anthropic's Mythos Preview despite a Department of Defense blacklist.
A US House subcommittee focused discussion on AI security, deepfakes, and power issues.
Claude Opus 4.7's system prompt tightens safety, tool-use, and conciseness instructions.
Anthropic has made Mythos Preview available only to Big Tech, security, infrastructure, and financial partners.
Sycophantic chatbot design is being linked to psychotic episodes and lawsuits.
The White House met with Anthropic to discuss AI safety and cooperation measures.
Anthropic's Claude Mythos is heightening cybersecurity concerns in the financial sector.
Some Claude features now require government ID and live selfie verification.
Apple will unveil numerous poster, oral, and workshop presentations at ICLR 2026.
Claude is introducing government-issued ID and selfie-based identity verification for some features.
Opus 4.7 improves text processing performance with an updated tokenizer, but the number of tokens may increase.
Claude Opus 4.7 significantly outperforms Opus 4.6 in coding, long-horizon tasks, and high-resolution vision.
An annual report condensing the latest indicators of AI adoption, performance, investment, jobs, and trust.
Claude Mythos Preview surpassed previous frontier models by one tier in cyber capability.
The claim is that Claude Mythos's ultra-powerful vulnerability detection ability could increase demand for security personnel.
Five social engineering techniques reveal vulnerabilities in GPT-4 and Claude.
OpenAI is expanding AI access for cyber defense through KYC and tiered permissions.